Rare diseases present unique challenges in the design of clinical trials due to limited patient populations, making it difficult to recruit enough participants to have potential clinical benefits also statistically significant. This thesis explores innovative trial designs, with a particular focus on the use of Bayesian methods and Master Protocols to address these challenges and enhance the research in this field. Bayesian methods offer several advantages for trials involving small sample sizes. By incorporating prior knowledge, enabling continuous updates, and supporting adaptive trial designs, the Bayesian framework provides flexibility and precision that is particularly valuable in the context of rare diseases. Additionally, Master Protocols, such as Basket and Platform trials, allow for the simultaneous evaluation of multiple treatments across different patient subgroups, thereby improving patient enrolment efficiency, optimizing resource allocation, and accelerating the process of evaluating novel therapies. This thesis has two main objectives. First, we aim to assess the robustness of three standard Bayesian methods, specifically adapted to continuous outcomes, in the analysis of basket trials within the context of rare diseases. The second objective, motivated by a context of rare diseases, is to develop a novel, sequential, single-arm Platform-Basket trial design. The results of the simulation study to assess the value of Bayesian models for the analysis of basket trials in rare diseases are very promising. In particular, when clear treatment effects are present within subpopulations, the three methods can be confidently applied even when sample sizes are very small (i.e. 4-7). However, when the treatment effects are mild or when there is a relevant heterogeneity in treatment responses, careful consideration of sample size and final cut-off probability for final decision is crucial during the planning phase of the trial. The novel two stage design we have proposed combines innovative elements to plan a complex clinical study involving a platform-basket trial back-bone, reinforced by the use of external information (e.g. from a completed trial on a similar disease) and an interim evaluation for futility. The use of interim analyses is impactful, enabling early decisions to either halt unpromising trials or expand the study to include additional subgroups based on predictive probabilities. We assessed the performance of this design through simulations that resulted sensitive to the choice of some parameters (e.g. prior weights, cut-off probability for the interim analysis and the final decision), even if it was in general robust when there are strong believes of highly effective treatments. It is important to emphasise that in the set-up of the protocol it is of paramount importance to study the properties of the proposed design through simulations. Indeed, the operating characteristics should be carefully evaluated by extensive clinical trial simulations whose results should be always part of the study protocol. In conclusion, the work presented here serves as a foundational starting point for enhancing our understanding of how to design a trial in the context of rare diseases and additional work is needed for a more comprehensive evaluation.
Le malattie rare presentano particolari sfide per quanto riguarda la progettazione di trial clinici a causa della limitata disponibilità di pazienti. Ciò rende difficile il reclutamento di un numero sufficiente di partecipanti allo studio che abbiano un potenziale beneficio clinico che sia anche statisticamente significativo. Questa tesi esplora dei disegni di studio innovativi, con focus particolare sull'uso dei metodi bayesiani e dei Master Protocols, per affrontare le difficoltà e contribuire alla ricerca in questo campo. I metodi bayesiani offrono diversi vantaggi per gli studi che coinvolgono campioni di piccole dimensioni. Incorporando conoscenze pregresse, permettendo aggiornamenti continui e supportando disegni adattivi, l’approccio bayesiano offre flessibilità e precisione, risultando particolarmente utile nel contesto delle malattie rare. Inoltre, i Master Protocols, come i Basket Trial e i Platform Trial, consentono di valutare contemporaneamente più trattamenti in diversi sottogruppi di pazienti, migliorando così l'efficienza nel reclutamento, ottimizzando l'allocazione delle risorse e accelerando il processo di valutazione di terapie nuove. Questa tesi ha due obiettivi. Il primo è quello di valutare la robustezza di tre metodi bayesiani standard, specificamente adattati alla presenza di endpoint continui, nell'analisi di un basket trial nel contesto di malattie rare. Il secondo obiettivo, motivato da un contesto di malattie rare, è quello di sviluppare un nuovo disegno di studio per un Platform-Basket trial, sequenziale, con il solo braccio di trattamento. I risultati dello studio di simulazione per valutare le performance dei modelli bayesiani nell'analisi dei basket trials in malattie rare sono molto promettenti. In particolare, in presenza di rilevanti effetti del trattamento nei vari sottogruppi di malattia, i tre metodi risultano già ottimali con campioni molto piccoli (ovvero 4-7 pazienti). Tuttavia, quando l'effetto del trattamento è lieve o c'è eterogeneità nelle risposte ai trattamenti, è fondamentale scegliere attentamente la numerosità campionaria e i cut-off decisionali durante la fase di pianificazione del trial. Il nuovo disegno a due fasi che abbiamo proposto combina elementi innovativi per progettare uno studio clinico complesso, utilizzando come base un trial Platform-Basket, rafforzato dall'uso di informazioni esterne (per esempio da un trial su una malattia simile già completato) e da una valutazione ad interim per futilità. L'uso delle analisi intermedie è particolarmente importante, in quanto consente di interrompere precocemente trial che non si dimostrano promettenti o di espandere lo studio includendo sottogruppi aggiuntivi, basandosi su delle probabilità predittive. Abbiamo valutato le performance di questo disegno attraverso simulazioni, che sono risultate sensibili alla scelta di alcuni parametri (come ad esempio i pesi scelti a priori, i cut-off per le decisioni ad interim e finale), anche se, in generale, il disegno è risultato robusto in presenza di forti evidenze sull'efficacia dei trattamenti. È importante sottolineare che nella definizione del protocollo è fondamentale studiare le proprietà del disegno che abbiamo proposto attraverso simulazioni. Infatti, le caratteristiche operative devono essere valutate attentamente tramite estese simulazioni di trial clinici i cui risultati devono sempre essere riportati nel protocollo di studio. In conclusione, questa tesi è solo il punto di partenza per migliorare la metodologia relativa alla pianificazione di studi clinici nel contesto delle malattie rare e ulteriori sviluppi saranno necessari per una valutazione più approfondita e completa.
RARE DISEASES: A COMPARISON OF BAYESIAN METHODS FOR BASKET TRIALS AND AN INNOVATIVE SEQUENTIAL DESIGN
RISCA, GIULIA
2025
Abstract
Rare diseases present unique challenges in the design of clinical trials due to limited patient populations, making it difficult to recruit enough participants to have potential clinical benefits also statistically significant. This thesis explores innovative trial designs, with a particular focus on the use of Bayesian methods and Master Protocols to address these challenges and enhance the research in this field. Bayesian methods offer several advantages for trials involving small sample sizes. By incorporating prior knowledge, enabling continuous updates, and supporting adaptive trial designs, the Bayesian framework provides flexibility and precision that is particularly valuable in the context of rare diseases. Additionally, Master Protocols, such as Basket and Platform trials, allow for the simultaneous evaluation of multiple treatments across different patient subgroups, thereby improving patient enrolment efficiency, optimizing resource allocation, and accelerating the process of evaluating novel therapies. This thesis has two main objectives. First, we aim to assess the robustness of three standard Bayesian methods, specifically adapted to continuous outcomes, in the analysis of basket trials within the context of rare diseases. The second objective, motivated by a context of rare diseases, is to develop a novel, sequential, single-arm Platform-Basket trial design. The results of the simulation study to assess the value of Bayesian models for the analysis of basket trials in rare diseases are very promising. In particular, when clear treatment effects are present within subpopulations, the three methods can be confidently applied even when sample sizes are very small (i.e. 4-7). However, when the treatment effects are mild or when there is a relevant heterogeneity in treatment responses, careful consideration of sample size and final cut-off probability for final decision is crucial during the planning phase of the trial. The novel two stage design we have proposed combines innovative elements to plan a complex clinical study involving a platform-basket trial back-bone, reinforced by the use of external information (e.g. from a completed trial on a similar disease) and an interim evaluation for futility. The use of interim analyses is impactful, enabling early decisions to either halt unpromising trials or expand the study to include additional subgroups based on predictive probabilities. We assessed the performance of this design through simulations that resulted sensitive to the choice of some parameters (e.g. prior weights, cut-off probability for the interim analysis and the final decision), even if it was in general robust when there are strong believes of highly effective treatments. It is important to emphasise that in the set-up of the protocol it is of paramount importance to study the properties of the proposed design through simulations. Indeed, the operating characteristics should be carefully evaluated by extensive clinical trial simulations whose results should be always part of the study protocol. In conclusion, the work presented here serves as a foundational starting point for enhancing our understanding of how to design a trial in the context of rare diseases and additional work is needed for a more comprehensive evaluation.File | Dimensione | Formato | |
---|---|---|---|
phd_unimib_891803.pdf
accesso aperto
Dimensione
14.58 MB
Formato
Adobe PDF
|
14.58 MB | Adobe PDF | Visualizza/Apri |
I documenti in UNITESI sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
https://hdl.handle.net/20.500.14242/195926
URN:NBN:IT:UNIMIB-195926